C4W1 Convolutional Neural Network Example - Typo/Error?

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

In the second slide, where all the shapes, size and parameters are summed up, I think there is an error with the activation shape for CONV1 and POOL1.

Could it be 28x28x6 instead of 28x28x8; and 14x14x6 instead of 14x14x8?

2

Hi Gong,

Thanks for your contribution.
There are many videos on week 1. Could you please be more specific in which video do you mean there is an error? I reviewed the first two videos of W1 and could not find it.

Best,

Rosa

3

Hi arosacastillo,

Thank you for your answer!

I think I defined pretty badly where the doubt was.

Reviewing my notes I see the question mark on the second slide of Convolutional neural network example. And as far as I remember, the doubt comes from the difference with the previous slide where the size of the CONV1 is 28x28x6 and in the summary table is 28x28x8.

Convolutional neural network example “lecture” that comes between Pooling layers and Why Convolutions? in Course 4.

Please find attached the two slides in hand.

Thank you in advance for your time.

IMG_1274
IMG_12741865Ă—1067 267 KB

IMG_1275
IMG_12751863Ă—1093 226 KB

1 Like
4

I have the same doubt; it seems that the correct number for both cases is 6 instead of 8 according to the slide, Conv 1 (28x28x6) and Pool 1 (14x14x6). What is the math equation to determine #parameters? I need this answer.

5

Hi,

Yes, now I see what you mean and indeed it looks like a typo. I checked also with the LeNet-5 architecture

So it should be
28x28x6 Activation size 4704
14x14x6 Activation size 1176

Will pass this to the DLS team.

Many thanks and sorry for the late response.

Happy learning,

Rosa